Books can document cultural biases from published times
Anne Taylor/Aramie
While artificial intelligence that picks up sexist and racist bias is a well-known and persistent issue, researchers are now turning this into benefits to analyse social attitudes throughout history. Training AI models on certain decade novels can instill them with biases from that era, providing new ways to study how cultural biases have evolved over time.
Large-scale language models (LLMs) such as ChatGpt learn by analyzing large collections of text. They tend to inherit biases found in training data.
Source: www.newscientist.com
